LK-99Gods will soon have our redemption
>>17049003Say it with me.PERMANENT UNDERCLASS IS COMING.No UBI. You will all starve to death.
>>17049003Sauce?
And just like that the race is over.https://openai.com/index/navier-stokes-solution/GG no remake. Mass suicides are coming.
Does this shit have any real world application beyond what we already know?
>solve millennium level problems>can't even write 3/10 twenty page story from scratch In 5 years you'll need to pivot to either becoming visionary artist or wiping old peoples ass.
>>17049003Is this shit actually real? Where is the Sabine video? What is the bullshit meter at?
https://openai.com/index/navier-stokes-solution/
>>17049070>>17049077So much copium.Years ago it was "AI would never hands". Now the goalposts have moved. Your jobs are being replaced.
>>17049083I'm not a stemfag I'm just here for answers
>>17049063>While unlikely, we cannot rule out that de-identified data derived from their usage of our products helpedimprove our models.So basically they're liars and stole from TB. This is why you dont put any of your research into the models.
>>17049086no it does not have real world applicationsit's purely to dunk on the entire math community
>>17049087There is no such thing as stealing.Science should always come first. We don't need human ego when you're already obsolete.
>>17049070Navier-stokes equations are extensively used for any simulation involving fluids or that can be interpreted as fluids. Aviation air flow and water simulation are good examples. But just like any other result out of non-data-driven mathematical or physical modeling, it's only really ever been a toy. Sometimes, a toy is all we have. That's all. So if openai's proof is correct, then it means we're forced to do navier-stokes the slow (current) way, whereas if they had proven the converse, it would have been feasible to find solutions that could solve navier-stokes very cheaply (computationally) for any chosen moment in time in the simulation.Really, all it shows is what we've already always known: that navier-stokes is a bad model.
>>17049117fluid dynamics moved on from basic navier stokes decades ago. it is (or was) mainly a theoretical concern. singularities don't exist in reality anyways.
>>17049123>singularities don't exist in reality anyways.Yes, that's the important part which is why I said that all it shows is that navier-stokes is a bad model. It really has no further implications, but we already knew that before.>fluid dynamics moved on from basic navier stokes decades ago.Depends on the application. Unless you mean things like RANS vs LES etc. but those are still NS.
~350 000 lines of lean code for the navier stokes parthttps://github.com/openai/NavierStokesAndEulermath is like this now
>>17049123Singularities dont exist in reality, but the question is regarding finite time blow up, which has practical applications. Interestingly this specific blow up aligns well with the observed phenomena of vortexing, which can cause air to be sucked into pumps. In hindsight I'm surprised this wasn't seen sooner.
We need a new set of millenium problems for AI which are "shit people actually care about">Problem 1: How to build a fusion reactor using currently existing materials that creates more power than you need to put in to start the reaction>Problem 2: How to produce a room temperature superconducting material with practical application for energy transport and the power grid>Problem 3: Defined as a less-destructive-than-chemo treatment to your pick of one of the most common forms of difficult to treat cancer which cuts mortality rate of the cancer to below 10%Feel free to add on anyone.
>>17049158>Problem 4: XX women with a functioning penis and sperm production
>>17049123>>17049149Which is to say that forced, finite-time blow up corresponds to where if you put enough energy into the system arbitrarily large vortices can form, which in this case engineers want to submerge pipes deep enough to prevent this vortexing from happening and sucking air into the water pump.>https://www.youtube.com/watch?v=T8jcNAPZ5isNormally this vortexing phenomena has to be determined computationally, but this solution could open up some analytic methods with which to compute how much pipes need to be submerged. Basically, this will have practical implications for engineering.
>>17049167You don't understand any of it.
>>17049158Problem 3 has been repeatedly solved in many cancers. The main issue is regulatory, not tech.
>>17049158Problem 4: How do I get gf
GUYS I JUST SWIRLED MY ONIONS LATTE AND IT EXPLODED
>I fucking love science>n-no don't solve it like thatwhy are they crying
https://cdn.openai.com/pdf/32d9f210-8b73-45e0-91bc-82a30aef8a9a/navier-stokes.pdf
>>17049176500 page proof and a million lines of Lean code that irrefutably demonstrate we're all gonna make it brehs
>>17049195> 166 pages of pure slopyeah no gonna read it if they cna't write it down in a standard article 5 pages max its probably shit. Glad I didn't study math.
>>17049088Disappointing. Call me when we solve P=NP I guess
>>17049291Solving this problem is a massive boost to scientific output since now big brains that have been stuck on this problem for 100 years can be used on other problems.
>>17049070>Does this shit have any real world application
call me when these AI agents solve aging or something like that
All the breakthroughs that are needed until AI is useableCold fusion Room temperature superconductorHomomorphic encryption Reversible computingThermodynamic computing
>>17049294Not at all
>>17049352The difference is all these things except AI were immediately useful, though niche.
>>17049411All of those took years before they were in any sort of practical use. Airplane took roughly a decade before it was in any sort of commercial use for instance. Comparatively ChatGPT got into commercial use in mere moments.
>>17049426Not at all true. Even for airplanes, the first commercial aircraft flew in 2014 but it was preceded by other kinds of crafts. Air travel was already known to be useful and common. Nobody claimed that airplanes at the time were magical transformative and would make horses obsolete as everyone would just use an aircraft for local transport.The very first computer was specifically built for a fixed purpose and achieved it as expected, the exact anti-thesis of AI. Cars are similar to the planes situation, but they were really shitty for a while. However, they could be used from day 0 for transport and did successfully transport. AI doesn't do anything it's claimed to do at even the most basic level.
>>17049411>all these things ... were immediately usefullmaooooooo
>>17049438>no argumentEvery time
>>17049003
>>17049442>i need people to refute every retarded statement i spoutA reminder that it takes then times as much energy to refute a retarded statement than it takes to spout one.
>>17049003i cant wait for all the migger coping and seething this is going to cause the entire fucking month.
>>17049449I accept your surrender.
>>17049003Wtf you called it. They're firing canonballs at each other now that the IPO is imminent
>>17049042Permanent underclass was already here, fool.
>>17049411How was a glorified glider immediately useful?
>>17049459>completely fake postKek
>>17049459I'm going to kill youIn Los AngelesAt your house
>>17049479I've seen your kicks Hasan Piker. I could pin you down with one hand and rape your bussy
>>17049467Just 2 years after the first non-propelled flight, planes were already used to travel between towns.
>>17049798>just 2 years after the first release of llms, llms were already used by students to do their homework for them
>>17049969Retarded or pretending?
>>17049990Anon, EVERYONE uses llmsThe few who refuse to touch them are the vast, vast minority
>>17050007Oh no... it's retarded :(
>>17050016Stay safe in your little bubble, anon.At some point driving cars will be illegal and you'll be forced to interact with LLMs.
>>17049087Aka you claim someone was hiding things that could have helped to solve this for themselves instead of sharing it? Its fucked up.
>>17049432>first commercial aircraft flew in 2014What?
>>17050047Brainfart.1914.
> tfw soon instead of fat you are getting... lean!> you all
Say I am getting a grant/a project/a PhD on X.Some asshole puts "solve X" in an LLM.My job is done, or the problem is extremly hard to solve and I fail.It is over for transparency and collaboration.
>>17049003>Navier-Stokes is solved>0 peer reviews of the whole thing>Stan Palasek checked the model they used and found a critical flaw>OpenAI states they aren't going after the Millenium prize cause it would mean it has po be published in a math journal and not get disproved in the next two yearsI say it's BS to grab more investors' money before "their" "solution" gets disproven in like two months
>>17049158The Chudnovski algorithm: Prove that it is the most effective way to compute pi or provide a new algorithm.
>>17050154so your bright mind can be put to something useful instead (such as accounting) and everyone wins
>>17050186The proof is formalized in Lean, anon. You cannot falsify a Lean program, otherwise it simply wouldn't run or resolve. It enforces mathematical correctness of whatever is written in the language.
>>17050242Yes you can and it's been done many times in the history of math. A quick google would save you the embarrassment of this ludicrous post.
>>17050255>Yes you can and it's been done many times in the history of math.To what part of my post are you responding, exactly? Because I thought the Lean language was a more recent invention. Did the ancient Arabs use Lean as well? How fascinating!
>>17050258>the history of math ended in 3000 BCSasuga
>>17050259It did? Wow, tell me more!
>>17050242LEAN only verifies the internal mathematical logic, it can't verify the whole proof.Mathematicians use LEAN as an intermediate step, so not to publish something internally contredicting
>>17049003This kills the scientist
>>17050279A bit ironic you'd use ChatGPT to generate a response, innit?
>>17050295I prefer to use my own intelegence
>>17050209Accountancy is dull, dull, dull!
>>17049042Permanent underclasses are the onlƴ thing constant throughout historƴ. Especially in Europe which is extremely feudal and ideological.
>>17049432>AI doesn't do anything it's claimed to do at even the most basic level.@grok is this true?
>>17049158>fusion reactorlisten. no matter what free energy source they find we won't enjoy any of it. in yurop we pay 70-80% taxes on gas. if gas was free for them we'd still pay almost the same.you are never ever getting away from taxes. UBI is opposite spectrum of taxes, and taxes aren't coming down.it's absolutely clear what will happen anon. you know it, I know it...everybody knows it.
>>17049411the problem with that observation is that life was made useless before AI.
>>17050337No wonder your spelling is shit.
>>17050683>everybody knows it.That U.B.I. Already Exists, Pasty?Gay pasty criggas from GodKnowsWhereP=NP Solved?
>>17050943>P=NP SolvedI tried with expounding and mathematical rounding and rounding deduction to outcome result instead of algebra. It was expanded and the entire expands treated somewhat algebraicly
>>17050685Rather, it's why there is so much money behind AI despite it being useless.
>>17050943>That U.B.I. Already Exists, Pasty?can exist as long as others are working and paying taxes. neet bux are basically other people's money, other working people's money
>>17049070Swap fluid with gravity profit with endless singularity gibs.
>be OpenAI>steal insight>claim breakthrough>ignore attribution>make thinly veiled threats>refuse to elaborateThose clowns are actively disbanding the humans' research collective. Unless they are removed, this will end well for our civilization.
>>17049373You cant cure aging. Immortality is forever out of reach
>>17051976I refuse.
>>17051976>>17051978i dunno man, if AGI becomes real, who knows what we can do?
>>17051976>Immortality is forever out of reachutter nonsense. but I agree bio immortality is pretty fucking retarded. makes things way worsepeople don't even understand what they actually want
>>17052066If not bio immortality, then what?
>>17052067you need save&resume anon. you do not understand what bio immortality does to one's brain.atm you have what, 30-60 years to lose at most, the older you grow the less fucks you give because not much loss anyway.but when you have thousands of years, or even billions, to lose, brother trust me, it fucks with your brain in ways you cannot even comprehend. the fear of dying from some random accident will transform you into something you wouldn't be able to now recognize lolit's the worst possible thing that can happen to us as a species. we will need a new word to define that level of paranoia
>>17052072>you do not understand what bio immortality does to one's brainhttps://youtu.be/VM5qi0aSstQ
>>17051976>You cant cure aging.lollmaoThere's a human trial going on right now that, if proven to work, is a pathway towards rejuvenating every type of cell in the body.And with future AI models being given access to biolabs, there's literally no way we won't invent some way to become immortal before the year 2050.My personal estimate? 2044. It took us 6 years to go from a barely useable chatbot to something that can functionally interact with the (digital) world better than most people can.Within the next 6 years we'll have self-improving AI, either due to a new architecture or due to existing AI being able to generate reality-based data and use it to create a new version of itself.Within the next 12 years we'll have AI that is capable of solving literally everything we humans can think of.Within the next 18 years, AI will be running most tech in the world, handling everything from traffic signs to self-driving cars to make-up R&D.
>>17052083imagine going to jail for 4000 years lol
>>17049042Seems you're getting ready to drink the kool aid
>>17052085An eternity for someone who expects to live to 80.An amusing moment for someone who expects to live forever.
>>17052083Even without AI replacing human researchers outright, it's already accelerating research simply by being endlessly available codemonkeys with expert domain knowledge, something that was a real bottleneck in bio and many other fields.
>>17052083how would you fix the memory problem? The brain can only store so many memories, and it's not really made for much beyond 100
>>17052107The brain takes care of that "problem" pretty well on its own, we also invented writing to deal with that state of affairs.
>>17052101not if you get shanked in there
>>17052114can't get shanked if you kill everyone thereyou have infinite time to do it as well
>>17052107>how would you fix the memory problem?Huh? What memory problem?
https://youtu.be/wt6nwvGJiN8
>>17052120that will just prolong your sentence, and new people will come in there again. so you have to do more killing, which will again increase your sentence.your plan seem kinda dumb. this is proof humans will hurt themselves with it, even more
>>17052122the fact that we only have a finite amount of neurons to make memories with
>>17052125Memories? What were we talking about again?
>>17052125I mean you only really need a few to remember the essential stuff innit. fuck does it matter if you forget about Rihanna? or like whatever, armadillos
>>17052128i wish i could forget about armadillos. imagine finding out they exist, it would blow my mind
>>17052123What's this "penshon" thing they're talking about?>>17052125Neurons lose and form connections all the time.If you want to keep remembering something, keep doing things that reinforces the connections.Sure, you'll eventually remember celebrating your 15th birthday at a zoo and all your classmates and history teacher were there while your parents got divorced in a bank where someone congratulated you on buying your first home? But I'm sure they'll invent some brain implant that'll take care of that.
>>17052131>What's this "penshon" thing they're talking about?that's the joke. they'd be working forever. literally
>>17052107That's a pubsci myth
>>17052105The opposite. Especially in biology you can't actually tell if the output is correct, so you don't see bugs until they hit the lab months down the road with dozens of thousands of dollars spent. P-hacking was already a widespread problem before, now it's a million times worse. You already couldn't trust compute-only results, now you can't even trust bio results because many are only interpretable by computers, and they're no longer using vetted pipelines anymore, just made-up ai bullshit.
>>17052144You sound like your exposure to modern AI is various variants of ChatGPT.
>>17052146Oh yeah? Well you sound like your mom and dad were related prior to marriage
>>17052146>sub-65-iq post with no argumentClassic AI psychosis
>>17052144All I can say is that not everyone is as severely retarded as your lab apparently is.
>>17052316My lab doesn't use any AI because they are not retarded like you and your pajeet friends.
>>17049070spaghetti rolling
